Correlation volcano - stjude/proteinpaint GitHub Wiki

The correlation volcano plot displays the correlation between a predefined set of variables (e.g. treatments, etc.) and a numeric term such as gene expression or phenotype value. The x-axis represents the correlation coefficient (e.g., Pearson’s r), while the y-axis displays the statistical significance as −log₁₀(p-value).

By default, red points indicate anticorrelation (negative correlation), and blue points indicate correlation (positive correlation). The radius of each data point is scaled according to the sample size used in the correlation calculation, ranging between the largest and smallest sample sizes in the dataset — allowing users to visually assess confidence in correlations.

How this Plot Works

The feature term (shown in the plot title) is plotted against pre-defined variable terms set by the dataset.

To calculate the correlation:

1. There must be more than 3 values for each variable.
2. The standard deviation for the values for both variables is greater than 0.05

If a variable is skipped for either reason above, the variable name will be shown to the right of the plot.

The plot is not available when its not supported by the dataset.

How to Launch

A 'Correlation volcano' button will appear in the 'Charts' tab of the portal UI. Choose a term from the pop-up menu to launch the plot.

Interactive Features

Users can launch the scatter plot and change the controls from the burger menu to configure the plot.

Launch Scatter Plot

Clicking on a data point opens a 2D scatter plot showing the individual-level relationship between the feature term and variable.

Controls

Users can change the correlation method (e.g., Pearson, Spearman), select the type of p-value, adjust significance thresholds, and configure other display options from the burger menu to the left of the plot.